Optimization of laser welding process parameters for super austenitic stainless steel using artificial neural networks and genetic algorithm

Penetration Depth Prediction of Infinity Shaped Laser Scanning Welding Based on Latin Hypercube Sampling and the Neuroevolution of Augmenting Topologies

This paper builds an infinity shaped (“∞”-shaped) laser scanning welding test platform based on a self-developed motion controller and galvanometer scanner control gateway, takes the autogenous bead-on-plate welding of 304SS with 3 mm thick specimens as the experimental objects, designs the experimental parameters by the Latin hypercube sampling method for obtaining different penetration depth welded joints, and presents a methodology based on the neuroevolution of augmenting topologies for predicting the penetration depth of “∞”-shaped laser scanning welding. Laser power, welding speed, scanning frequency, and scanning amplitude are set as the input parameters of the model, and welding depth (WD) as the output parameter of the model. The model can accurately reflect the nonlinear relationship between the main welding parameters and WD by validation. Moreover, the normalized root mean square error (NRMSE) of the welding depth is about 6.2%. On the whole, the proposed methodology and model can be employed for guiding the actual work in the main process parameters’ preliminary selection and lay the foundation for the study of penetration morphology control of “∞”-shaped laser scanning welding.

Multi-Objective Optimization of Resistance Welding Process of GF/PP Composites

Thermoplastic composites (TPCs) are promising materials for aerospace, transportation, shipbuilding, and civil use owing to their lightweight, rapid prototyping, reprocessing, and environmental recycling advantages. The connection assemblies of TPCs components are crucial to their application; compared with traditional mechanical joints and adhesive connections, fusion connections are more promising, particularly resistance welding. This study aims to investigate the effects of process control parameters, including welding current, time, and pressure, for optimization of resistance welding based on glass fiber-reinforced polypropylene (GF/PP) TPCs and a stainless-steel mesh heating element. A self-designed resistance-welding equipment suitable for the resistance welding process of GF/PP TPCs was manufactured. GF/PP laminates are fabricated using a hot press, and their mechanical properties were evaluated. The resistance distribution of the heating elements was assessed to conform with a normal distribution. Tensile shear experiments were designed and conducted using the Taguchi method to evaluate and predict process factor effects on the lap shear strength (LSS) of GF/PP based on signal-to-noise ratio (S/N) and analysis of variance. The results show that current is the main factor affecting resistance welding quality. The optimal process parameters are a current of 12.5 A, pressure of 2.5 MPa, and time of 540 s. The experimental LSS under the optimized parameters is 12.186 MPa, which has a 6.76% error compared with the result predicted based on the S/N.

Nd: YAG Pulsed Laser Dissimilar Welding of UNS S32750 Duplex with 316L Austenitic Stainless Steel

Duplex stainless steels (DSSs), a particular category of stainless steels, are employed in all kinds of industrial applications where excellent corrosion resistance and high strength are necessary. These good properties are provided by their biphasic microstructure, consisting of ferrite and austenite in almost equal volume fractions of phases. In the present work, Nd: YAG pulsed laser dissimilar welding of UNS S32750 super duplex stainless steel (SDSS) with 316L austenitic stainless steel (ASS), with different heat inputs, was investigated. The results showed that the fusion zone microstructure observed consisted of a ferrite matrix with grain boundary austenite (GBA), Widmanstätten austenite (WA) and intragranular austenite (IA), with the same proportion of ferrite and austenite phases. Changes in the heat input (between 45, 90 and 120 J/mm) did not significantly affect the ferrite/austenite phase balance and the microhardness in the fusion zone.

Prediction of Bead Geometry Using a Two-Stage SVM–ANN Algorithm for Automated Tungsten Inert Gas (TIG) Welds

Prediction of weld bead geometry is critical for any welding process, since several mechanical properties of the weldment depend on this. Researchers have used artificial neural networks (ANNs) to predict the bead geometry based on the input parameters for a welding process; however, the number of hidden layers used in these ANNs are limited to one due to the small amount of data usually available through experiments. This results in a reduction in the accuracy of prediction. Such ANNs are also incapable of capturing sudden changes in the input–output trends; for example, where a wide range of heat inputs results in flat crown (zero crown height), but any further reduction in the current sharply increases the crown height. In this study, it was found that above mentioned issues can be resolved on using a two-stage algorithm consisting of support vector machine (SVM) and an ANN. The two-stage SVM–ANN algorithm significantly improved the accuracy of prediction and could be used as a replacement for the multiple hidden layer ANN, without requiring additional data for training. The improvement in prediction was evident near regions of sudden changes in the input–output correlation and can lead to a better prediction of mechanical properties.